Area source radon diffusion model of virtual-point source method based on Trust Region strategy optimization
When the area source Gaussian diffusion model is used to simulate the radon diffusion concentration of nuclides in uranium tailings ponds,the finite segment area source model is complex to calculate,and the simulation results of the virtual-point source model are neither accurate nor conservative.In response to the above problem,a function and parameter optimization method based on trust region algorithm has been proposed.Constructing a set of nonlinear functions that approximate virtual-point source model,and defining the error between simulated values of finite segment area source model and each function in the function library as the objective function.Then the function library will be called under the trust region strategy and iteratively optimized till the optimal function and its optimal parameter value are output.Through the comparison of concentration distribution and fitting test,it indicates that the simulated value near the source of the optimization function is slightly higher than the counterpart of finite segment area source model,which proves the optimization function partly has a conservatism;the goodness-of-fit near the source between the simulated values is higher than 0.91,and higher than 0.98 in most areas,showing that the simulation results are reliable.In addition,according to the analysis of the influencing factors of error,the optimization function can adapt to the changes of other parameters when the source area and atmospheric stability remain unchanged,which greatly simplifies the calculation process of the actual radon concentration assessment.
Gaussian plume modeluranium tailings pondvirtual-point source modeltrust region algorithm
国家科技期刊平台
NSTL
万方数据
